TSE‐639

“Anenvironmentaltaxtowardsmoresustainablefoodconsumption:empiricalevidenceoftheFrenchmeatand

marinefoodconsumption”

CélineBonnet,ZohraBouamra‐MechemacheandTifennCorre

April2016

An environmental tax towards more sustainable food

consumption: empirical evidence of the French meat and

marine food consumption

Celine Bonnet*, Zohra Bouamra-Mechemache�and Tifenn Corre�

April 12, 2016

Abstract

After fossil fuels, agricultural production and fisheries are industries with thelargest impact on the environment in terms of greenhouse gas (GHG) emissions,especially in the production of ruminant meats such as beef, veal or lamb. In orderto reduce this environmental impact, consumers can change their food consump-tion habits to utilize less polluting products such as white meats or vegetable foodproducts. We analyze whether or not a CO2 equivalent (CO2-eq) tax policy canchange consumer habits with respect to meat and marine purchases, and usingdifferent indicators, we examine the effect of such a tax policy on the environ-ment. We also infer the implications of such a tax on nutritional indicators as wellas on consumer welfare. First, to evaluate the impact of a variation in the priceof meat and marine products on consumption, we estimate a random coefficientslogit demand model using purchase data from the French household panel KantarWorldpanel. We define 28 meat and marine products, and divide them into eightmeat and marine product categories. This model allows us to estimate flexibleown- and cross-price elasticities of meat and marine products’ demand. Resultson the consumer purchase behavior model suggest that the demands for theseproducts are fairly inelastic, and substitutions occur both within and betweencategories for all products. Moreover, using two levels of a CO2-eq tax (e56 ande200 per tonne of CO2-eq per kilogram of product) applied to either all meatand marine products, only ruminant meats, or only beef, we show that a tax ofe56 leads to a very small change in GHG emissions, even if all meat and marineproducts are taxed. The most efficient scenario would be to tax only the beefcategory at a high level since it would allow a 70% reduction in the total variationof GHG emissions, and would be responsible for only 20% of the consumer welfaredamages generated when all products are taxed.

Key words: meat, demand analysis, environment, greenhouse gas, CO2-eq tax, con-sumer diet

*Toulouse School of Economics (INRA), 21 Allee de Brienne, F-31000 Toulouse France, Tel: +33(0)5 61 12 85 91, cbonnet@toulouse.inra.fr

�Toulouse School of Economics (INRA), 21 Allee de Brienne, F-31000 Toulouse France, Tel: +33(0)5 61 12 86 84, bouamra@toulouse.inra.fr

�Toulouse School of Economics (INRA), 21 Allee de Brienne,F-31000 Toulouse France, Tel: +33(0)5 67 73 27 81, tifenn.corre@toulouse.inra.fr

1

1 Introduction

After fossil fuels, agricultural production and fisheries are the two main industries thatimpact the environment the most. They are significant contributors in terms of climatechange but also in terms of eutrophication, land use, water use and toxicity. Amongagricultural activities, meat and dairy are two of the major impacting sectors, as alarge proportion of crops are indirectly used for the production of meat and marineproducts, resulting in a high land use (UNEP, 2010). The environmental impact offood is in large part determined by household diet and consumption habits. For in-stance, Reynolds et al. (2015) evaluate that meat and bakery products/flour/cerealsare the food categories with the largest footprint contribution for all household incomeclasses in Australia, and that the higher the household income, the larger the environ-mental burden.

Population and economic growth should lead to an even higher environmental im-pact on the future if patterns of production and consumption are not changed. More-over, as the aggregate world meat consumption, as well as per capita consumption,has increased over time (by 60% and 25% respectively between 1990 and 2009 fromHenchion et al. (2014)), this trend is predicted to continue in the future. For instance,the study of Fiala (2008) shows that if current consumption patterns continue, totalmeat consumption will increase by 72% between the years 2000 and 2030, mostly leadby a large increase in chicken and pork consumption. Such a trend is also observedin Europe. For instance, the share of meat in the Spanish diet increased between1970 and 2005 with an average annual meat consumption per capita that rose from11.7 kg to 65 kg (Rios-Nunez et al., 2013). This trend in the consumption per capitain the European Union (EU) is expected to be positive for all meats (except sheepmeat) between 2013 and 2022 with a decrease in the share of red meat in the totalmeat consumption in favor of white meat (Henchion et al., 2014). In France, the totalconsumption of meat has decreased since 1998. However, France still represents one ofthe three countries with the highest consumption of animal proteins in the EuropeanUnion (source FAOSTAT).1 In 2014, the quantity consumed in France accounts respec-tively for more than 20% of the total EU consumption (in tonnes equivalent carcass)for beef, veal and marine products, 14% for poultry and 11% for pork (FranceAgriMer,2015).

At the consumption end, some studies have analyzed how changes in consumptionhabits and diet may mitigate the environmental impact of food consumption. However,as shown in Hedenus et al. (2014), the literature on the mitigation potential throughdietary changes under the constraints of consumer preferences is relatively poor. Hede-nus et al. (2014) considered different assumptions on food consumption patterns usingFAO projection and two assumptions with respect to consumer preferences: “75% ofruminant meat and dairy products are replaced by other meat (on kcal basis)”, and“75% of animal food is replaced by pulses and cereals (on kcal basis)”. They concludethat environmental impacts can be mitigated only with dietary changes in which theconsumption of animal products is reduced. In the same vein, Tukker et al. (2011)estimate the impact of three simulated diet patterns (a pattern according to universaldietary recommendations, the same pattern with reduced meat consumption, and a

1http://faostat3.fao.org/

2

‘Mediterranean’ pattern with reduced meat consumption) with respect to a status quoscenario. They found a reduction of up to 8% in the environmental impact in reducedmeat scenarios but that higher exports will compensate for losses on the domestic meatmarket. In addition, as emphasized by McMichael et al. (2007) and Horrigan et al.(2002), the growth in meat consumption, and in animal fat in particular, can alsoincrease the risk of chronic diseases, and thus does not only exacerbate environmentalproblems but also health risks. Along these lines, using simulations of scenarios repre-senting different variants of meat consumption in Sweden, Hallstrom et al. (2014) showthe existence of beneficial synergies of a reduction in meat consumption in Sweden interms of health, greenhouse gas emissions and land use.

Changing consumption habits towards a more sustainable direction and achievinga reduction in meat consumption may be a difficult task even if it is part of the sus-tainability public policy objectives (Austgulen, 2014). In this paper, we propose toanalyse the impact of environmental price policies that target meat and marine prod-ucts based on the analysis of French consumers’ purchasing behaviors. This analysisrequires a precise knowledge of consumer demand for meat and marine products. Inthe literature focusing on animal products demand estimation, most studies use dataaggregated at the country or regional level, while only a few studies are conducted atthe individual consumer level. Moreover, most of the literature deals with the demandfor meat in North America (Gallet, 2010). In this paper, we develop a demand modelof French meat and marine products consumption. We develop a random coefficientslogit model using individual data from a French household panel that gives detailedinformation on food purchases. This discrete choice model allows for the analysis ofconsumer preferences for all purchase alternatives available in the market. This modelalso allows for the estimation of consumption patterns between the different meat andmarine products but also for substitution with an alternative food product aggregatecomposed only of vegetable food products. As far as we know, the proposed demandmodel is one of the most disaggregated ones with 28 possible meat and marine prod-uct alternatives proposed to consumers. This disaggregated model allows us to catchthe substitution pattern at a very precise product level. Given the demand patternsfor the different animal product categories, we analyze whether or not public policytools can be used to encourage more sustainable food consumption habits. We focuson environment-based taxes. Such taxes are used, for instance, in Sweden to reduceemissions (a carbon tax but also a sulfur tax). We focus on a GHG tax based on CO2

equivalent and examine if such a tool can efficiently guide consumers’ choice of foodconsumption (cf. Vinnari and Tapio (2012), Wirsenius et al. (2011) and Edjabou andSmed (2013)). We investigate the impact of different carbon tax policies.2 In 2007,the European Union committed to target a limitation of global warming to 2°C (asper pre-industrial times) in order to halve the global emissions by 2050. In order toachieve these objectives, GHG emissions should be reduced by 20% by 2020 at the1990 levels and by 60% by 2050, and the recommended carbon price should be set ate56 per ton of CO2-eq in 2020 and e200 per ton of CO2-eq in 2050 (Quinet, 2009).We use these levels of carbon prices to simulate the impact of an CO2-eq tax policyon meat and marine products. We compare the effects of taxing all meat and marineproducts given their contribution to climate change, only ruminant products, or onlybeef products, all of which have the highest environmental impact. We compare the

2The term ”carbon” in this paper will refer to GHG emissions or CO2-eq emissions.

3

results according to their environmental impacts as well as their effect on consumerwelfare. Finally, we infer the nutritional impacts of such policies in order to evaluatecorroborating results or not.

Our results on consumer purchase behavior suggest that the demands for thoseproducts are fairly inelastic. Own-price elasticities for categories of meat and marineproducts range from −0.82% to −1.07%. Substitutions occur both within and be-tween categories for all products. Moreover, using two levels of carbon tax (e56 ande200 per tonne of CO2-eq per kilogram of product) applied to either all the meatand marine products, only ruminant meats, or only beef, we show that a tax of e56leads to a very small change in GHG emissions, even if all meat and marine prod-ucts are taxed (−1.54%). Even a high level of e200 per tonne of CO2-eq is not asufficient level to meet the 20% objective threshold of GHG emissions reduction for2020, since it would only lead to a 5% decrease. Despite the weak effect of such atax, the most efficient scenario would be to tax the beef category only at a high level,since it would allow a 70% reduction in the total variation of GHG emissions but wouldonly generate 20% of consumer welfare damage generated when all products are taxed.

The following section motivates the use of taxes based on GHG emissions to miti-gate environmental emissions. Section 3 discusses the market for meat and fisheries inFrance. In Section 3, we also present our estimation methodology to estimate the de-mand for the different categories of meat and marine products, as well as the demandestimation results that drive the demand substitution patterns for these products inFrance. Section 4 presents the different CO2-eq tax policy simulations and analyzestheir impact on different environmental and nutritional indicators. The last sectionconcludes.

2 Motivation for an environmental tax policy and related literature

Market failures that lead consumers to make suboptimal decisions are one of the mainreasons to justify public intervention. Suboptimal food choices result from consumers’lack of information about the environmental impact of the food products on the onehand, but also by the externalities of such choices on wildlife, global pollution and onhuman health, on the other hand.

In order to change consumption patterns, different policy tools can be used in-cluding tax policies, information intervention programs or subsidies. Informationalmeasures have been analyzed in the literature mainly under the form of dietary recom-mendations, promotions via social marketing campaigns, labeling regulation, and/oreducational measures. While such tools clearly modify attitudes and behaviors towardshealthy diets, they do not seem to have a significant impact on consumers’ consump-tion, at least in the short/medium term (for a survey, see the report of Traill et al.(2013)). Environmental subsidies could be an option if they provide incentives to in-vest in environmental innovations. However, they are costly, as all taxpayers will haveto pay for the subsidies whatever their consumption of meat and marine products.Moreover, these subsidies are not able to drastically change consumers’ purchasingbehavior as consumers will not have to pay a higher price for high impacting meat andmarine products. Thus, among the available instruments, taxes are the most efficient

4

tools, as they directly address the negative externalities linked with environmentaldamage. The resulting prices can integrate environmental cost impacts such that bothconsumers and firms will adapt their behaviors to reduce their environmental foot-print. Thus, this policy instrument is that which is recommended by most economists(Quinet, 2009).

A tax can be implemented either directly on emissions, on the product input at theorigin of the environmental impact or on the final product purchased by consumers.From economic theory we know that it is more efficient to use a tax that directlytargets the source of the market failure. However, as highlighted by Edjabou andSmed (2013) and Wirsenius et al. (2011), in the case of agricultural products, themonitoring costs are high, the technical potential for emission reduction is low and thepossible output substitution exists such that emission or input taxes are less effectivethan output based taxes. Thus, we propose in this paper to estimate whether or notconsumption taxes differentiated by the level of GHG emissions of meat and marineproducts can mitigate environmental indicators. This idea of taxing animal productsis not new, as acknowledged by Vinnari and Tapio (2012). High enough taxes can beefficient tools to guide consumer decision-making. However the effectiveness of suchtaxes has not yet been fully investigated.

Few studies have explored the impact of an environmental tax on food consump-tion, most of which consider taxes based on CO2 emissions even if a multi-GHG taxcan be more efficient than a CO2 tax (Feng et al. (2010)). Edjabou and Smed (2013)analyze the impact of a tax in Denmark based on CO2 emissions on more than 20 foodproducts differentiated with respect to average GHG emissions. This analysis is basedon elasticity estimates from Smed et al. (2007). Their most efficient scenario leadsto a decrease in GHG emissions for an average household by 2.3%–8.8% (at a cost of0.15–1.73 DKK per kg CO2 equivalent), and their most effective scenario in reducingthe environmental footprint leads to a larger decrease in the GHG emissions by 10.4%–19.4% but at a higher cost (3.53–6.90 DKK per kg CO2 equivalent). Wirsenius et al.(2011) focus on GHG weighted consumption taxes on animal food products in the EU,based on a model of food consumption in the EU. They show that agricultural emis-sions in the EU27 can be reduced by approximately 32 million tons of CO2-eq with atax of e60 per ton CO2-eq, and that most of the effect of a GHG based tax on animalfood can be captured by taxing the consumption of ruminant meat alone. The studiesof Edjabou and Smed (2013) and Wirsenius et al. (2011) rely on demand elasticityvalues at a meat category aggregate (pork, beef, etc.). We contribute to this literatureby providing an exhaustive analysis of the demand for meat and marine products ata very detailed product level using a unique dataset. This allows for the estimation ofelasticities that can be comparable at a very disaggregated level, and for substitutionestimates according to product by product substitution, not only at a meat categorylevel.

Two different tax strategies could be used to reduce the environmental impact ofmeat and marine product consumption: first, reduce the global consumption of thoseproducts; and second, favor the substitution within the meat and marine categorytoward less impacting animal products. Analyzing the benefits of public policies thattarget this second option requires having precise knowledge of consumer demand for

5

such animal products. In this paper, we thus propose to analyze the impact of envi-ronmental price policies that target either all meat and marine products or some meatproducts based on the analysis of consumer purchasing behaviors towards meat andmarine products.

3 Consumption substitution patterns in the French meat and marinemarket

We use a structural model of demand to estimate consumer preferences for meat andmarine products and derive the substitution between these products.

3.1 Data on the French meat and marine market

To conduct the analysis on the consumption of meat and marine products, we useconsumer panel data collected by Kantar WorldPanel, a French representative surveyof over 25,766 households conducted in 2010. This database records information aboutall purchases of food products for each household in the panel (for example quantity,price, brand, characteristics of goods, and the retailer from which the products arepurchased). The data cover household purchases for home consumption. Out-of-homeconsumption, which represents around 25% of the total food consumption, is not takeninto account.

We aggregate the meat and marine purchases at a monthly level. A period is de-fined as four consecutive weeks of purchases, with the study carried out over 13 periods.We consider eight categories of meat and marine products (pork, beef, chicken, otherpoultry, other meats, marine food, eggs and ready-made meals), with each categorydisaggregated into several products (cf. Table 1). In particular, we choose to separatefresh meat from processed meat and to disaggregate these further according to the useof pieces. This is why the pork category contains more products than beef or chicken,for example. When products from the same category have low frequencies, they aregrouped into an ”other” aggregate.

We define an outside good, which represents a substitute for the 28 meat and ma-rine products. It is composed of vegetable food products, such as vegetables, starches,pulses and vegetable ready-made meals. The share of this vegetable food productis roughly 45% of the defined market. Defining such an outside option allows us toconsider substitutions between meat and marine food, and vegetable food. From ourdata, we compute that the average consumption of meat and marine food is about84g and 24g a day per person respectively; values consistent with the INCA2 report(Afssa, 2009).3 Pork meat and marine food products dominate the meat and marineproducts market since they represent on average 34% and 16% of meat and marineproduct purchases respectively (19% and 9% of the entire market). It is important to

3INCA2 (”Etude Individuelle Nationale des Consommations Alimentaires 2006”) is a survey onindividual food consumption over a seven-day time period for a sample of 2,624 French adults and1,455 children.

6

Table 1: Descriptive statistics

Product Market share (%) Price (e/Kg)

Fresh pork 1.34 (0.09) 6.38 (0.41)Cooked ham and roasts 5.60 (0.21) 10.57 (0.08)Fresh sausages 1.17 (0.30) 7.58 (0.26)Other sausages (Frankfurt, Strasbourg) 1.22 (0.10) 5.19 (0.15)Bacon 2.23 (0.23) 6.95 (0.12)Blood pudding 0.64 (0.16) 9.10 (0.28)Other pork (dry sausage, pate, processed pork) 6.77 (0.36) 11.15 (0.36)

Pork category 18.96 (0.66) 8.80 (0.21)

Minced beef 2.04 (0.16) 7.43 (0.13)Beef for grilling 1.37 (0.07) 14.93 (0.38)Other beef (for braising or boiling, marinated) 0.86 (0.14) 8.39 (0.85)

Beef category 4.27 (0.27) 10.37 (0.55)

Ready to cook chicken 0.35 (0.04) 4.98 (0.17)Chicken parts 0.99 (0.09) 6.57 (0.16)Other chicken (ham, breaded chicken) 1.79 (0.18) 8.50 (0.20)

Chicken category 3.12 (0.29) 6.52 (0.17)

Turkey 1.71 (0.10) 7.92 (0.22)Duck 0.81 (0.32) 17.26 (3.32)Other poultry (goose, ostrich, rooster) 0.66 (0.10) 7.89 (0.83)

Other poultry category 3.19 (0.35) 9.66 (1.27)

Veal 0.85 (0.09) 14.41 (0.57)Lamb, sheep 0.43 (0.06) 11.87 (0.62)Mixed meats 0.32 (0.03) 7.41 (0.55)Rabbit 0.19 (0.02) 8.67 (0.27)Other meats (horsemeat, game) 0.30 (0.11) 13.15 (1.47)

Other meat category 2.09 (0.12) 11.91 (0.39)

Fresh fish 2.13 (0.16) 12.43 (0.42)Fresh shellfish 0.80 (0.20) 7.98 (1.04)Processed fish and shellfish 6.09 (0.60) 10.71 (1.18)

Fish category 9.02 (0.54) 10.62 (0.58)

Eggs 3.51 (0.14) 4.96 (0.08)

Pizzas, quiches 0.99 (0.05) 6.57 (0.08)Snacking 1.58 (0.06) 6.28 (0.40)Other ready-made meals 8.44 (0.80) 6.84 (0.36)

Ready-made meal category 11.01 (0.80) 6.73 (0.32)

Outside Good 44.83 (1.26) -

(.) : standard deviation

Numbers on grey rows are weighted means

7

note that ready-made meals make a large part of consumer consumption, represent-ing 11% of the purchases. According to the INCA2 report, there was no change inthe number of ready-made meals between 1989 and 2006. Among meat and marineproducts, beef (mainly for grilling), marine food products, and some specific poultryproducts such as duck (e17 per kilogram), and other meats such as veal, are the mostexpensive products, with a weighted average price around e10 per kilogram or more.Households face lower prices for eggs, chicken, ready-made meals and pork (except forsome products such as cooked ham and roasts), which would explain the bigger marketshares for these categories. Prices per category tend to be stable within the consideredperiod as the standard deviations of prices (indicating the variability of prices overperiods) are small for the meat categories. However, the variability is larger at theproduct level within meat categories, especially for duck, other meats and processedmarine food products.

3.2 Demand model

We use a random utility approach, in particular, a random coefficients logit model, toestimate the demand model and the related price elasticities. This structural modelbased on consumers’ utility generates flexible substitution patterns of consumers (Rev-elt and Train, 1998). We assume that the indirect utility function Vijt for consumer ibuying product j at period t is given by:

Vijt = βj + αijpjt + εijt, (1)

where βj is a product fixed effect that captures the (time invariant) unobserved prod-ucts characteristics and then represents the average preference of consumers for productj, pjt is the price of product j at period t, αij is the marginal disutility of the price forconsumer i, and εijt is an unobserved individual error term. We assume that αij variesacross consumers because their disutility with respect to prices could be heterogeneous.We assume that the parameter has the following specification:

αij =C∑c=1

αcj + σνi,

where αcj is the mean price sensitivity for consumers buying a product in a category c,C is the total number of categories, νi captures unobserved consumer characteristics,and σ measures the dispersion of the unobserved heterogeneity from the mean pricesensitivity. We assume a parametric distribution for νi denoted by Pν(.) and Pν isindependently and identically distributed as a standard normal distribution.

We can then break down the indirect utility into a mean utility:

δjt = βj +

C∑c=1

αcjpjt,

and a deviation from this mean utility µijt = pjtσνi. The indirect utility is then givenby:

Vijt = δjt + µijt + εijt.

8

The consumer can decide not to choose one of the considered products. Thus, we intro-duce an outside option that allows for substitution between the considered meat andmarine products, and a substitute with vegetable food. The utility of the outside goodis normalised to zero. The indirect utility of choosing the outside good is Vi0t = εi0t.

Assuming that εijt is independently and identically distributed like an extreme valuetype I distribution, we are able to write the individual probability for consumer i tobuy product j at time t in the following way:

sijt =exp (δjt + µijt)

1 +J∑k=1

exp (δkt + µikt)

.

The aggregated market share of product j at period t is then given by (Nevo, 2001):

sjt =

∫Ajt

exp (δjt + µijt)

1 +J∑k=1

exp (δkt + µikt)

dPν(ν),

where Ajt is the set of consumers who have the highest utility for product j at periodt, a consumer being defined by the vector (νi,εi0t,...,εiJt ).

The random coefficients logit model generates a flexible pattern of substitution betweenproducts, driven by the different consumer price disutilities αij . Thus, the own- andcross-price elasticities of the market share sjt can be written as:

∂sjt∂pkt

pktsjt

=

{−pjtsjt

∫αijsijt(1− sijt)φ(νi)dνi if j = k

pktsjt

∫αiksijtsiktφ(νi)dνi otherwise.4

Own-price (cross-price respectively) elasticity of market share sjt represents the vari-ation of market share sjt when the price of product j (k different from j respectively)at time t increases by 1%.

3.3 Price elasticities

We estimate the demand model (1) using individual data.5 From the demand estima-tion results and market share estimates, own- and cross-price elasticities of demandamong meat and marine products can be computed. From these individual productelasticities, we also compute aggregated own- and cross-price elasticities by meat andmarine product categories.

Table 2 reports the own-price elasticities at the product level in the second col-umn, the cross-price elasticities within the same meat or marine product category inthe third column and the cross-price elasticities between products of different product

4Where φ is the probability density function of the normal distribution.5We use a subsample of 500,000 over 7,260,307 observations to estimate the demand parameters.

This subsample is representative of the whole sample in terms of purchase shares over products andtime periods. The results are available upon request.

9

categories in the fourth column. The own-price elasticities vary from −0.87 to −1.11,meaning that the demands at the product level are rather inelastic. Consumers aremore responsive to changes in the own-price of fresh pork and sausages, fresh shellfish,pizzas, and snack food, and less responsive to changes in the own-price of duck, vealand other meats. The last three products are the more expensive products, mean-ing that those high-quality level product demands are the most inelastic. Table 2 alsoshows that substitutions occur both within and between categories for all products. Formost of the products, these substitutions are higher towards other products from thesame product category. However, this is not the case for minced beef and other beef,two of the three chicken products, turkey, lamb/sheep and mixed meats for which thecross-price elasticities are higher towards other products from other meat categories,and for other pork sausages where the substitution with vegetable food product is thehighest.

At the aggregated category level, own-price elasticities of the eight categories arebetween −0.82 and −1.07 (Table 3). These estimates are consistent with other stud-ies on meat consumption performed at the category level. From the meta-analysisconducted in Gallet (2010), predicted price elasticities range from −0.78 (poultry) to−1.62 (lamb) when considering studies from different countries around the world. Thiselasticity is evaluated at −0.85 at the world meat level and −0.88 for Southern Eu-rope, while the demand for meat is more elastic in Northern Europe with an elasticityevaluated at −1.016. A recent study conducted by Dong et al. (2015) on US householdpurchases confirms the inelasticity of meat demand in the US. Based on a censoredAIDS demand system model, they found estimates varying from −0.48 (seafood prod-ucts) and −0.75 (ground beef).

10

Table 2: Intra and inter category cross-price elasticities for each meat product

Cross-priceOwn-price elasticitieselasticities Same Category Other Categories

OGOther Products Other Products

Fresh pork -1.1023 0.0161 0.0145 0.0149Cooked ham and roasts -0.9940 0.0873 0.0771 0.0351Fresh sausages -1.1003 0.0154 0.0138 0.0108Other sausages (Strasbourg) -1.0937 0.0125 0.0113 0.0156Bacon -1.0873 0.0280 0.0255 0.0229Blood pudding -1.0889 0.0094 0.0084 0.0049Other pork -0.9703 0.1069 0.0932 0.0388Minced beef -1.0529 0.0208 0.0292 0.0137Beef for grilling -0.9408 0.0223 0.0210 0.0032Other beef -1.0507 0.0103 0.0118 0.0047Ready to cook chicken -1.0062 0.0037 0.0052 0.0013Chicken parts -0.9509 0.0130 0.0148 0.0024Other chicken -0.9049 0.0298 0.0268 0.0031Turkey -1.0062 0.0182 0.0258 0.0075Duck -0.8674 0.0223 0.0129 0.0008Other poultry -0.9078 0.0146 0.0128 0.0013Veal -0.8836 0.0093 0.0063 0.0004Lamb, sheep -1.0231 0.0073 0.0092 0.0027Mixed meats -0.9114 0.0042 0.0047 0.0004Rabbit -0.9011 0.0041 0.0038 0.0003Other meats -0.8829 0.0078 0.0049 0.0003Fresh fish -1.0520 0.0339 0.0291 0.0141Fresh shellfish -1.1101 0.0096 0.0084 0.0084Processed fish and shellfish -1.0112 0.0896 0.0790 0.0501Eggs -1.0668 . 0.0407 0.0352Pizzas, quiches -1.0914 0.0145 0.0124 0.0079Snacking -1.0872 0.0230 0.0198 0.0137Other ready-made meals -0.9739 0.1325 0.1104 0.0665

11

Tab

le3:

Ow

n-

and

cros

s-p

rice

elas

tici

ties

Pork

Beef

Ch

icken

Oth

er

Oth

er

Fis

h,

Eggs

Read

y-m

ad

eO

Gp

ou

ltry

meat

shellfi

shm

eals

Pork

-0.8

153

0.2

185

0.13

110.

1800

0.05

780.

2829

0.315

20.2

836

0.1

431

Beef

0.0

649

-0.9

825

0.04

170.

0509

0.02

240.

0665

0.069

60.0

669

0.0

216

Ch

icken

0.0

475

0.0

488

-0.9

059

0.04

810.

0311

0.047

40.0

443

0.0

477

0.0

067

Oth

er

pou

ltry

0.0

494

0.0

465

0.03

81-0

.9589

0.02

330.

0501

0.050

00.0

504

0.0

115

Oth

er

meats

0.0

313

0.0

379

0.04

700.

0419

-0.8

482

0.03

010.

025

50.0

302

0.0

022

Fis

h,

shellfi

sh0.1

286

0.1

007

0.05

750.

0815

0.02

41-0

.9659

0.1

518

0.1

347

0.0

727

Eggs

0.0

452

0.0

333

0.01

660.

0256

0.00

630.

0475

-1.0

668

0.0

475

0.0

352

Read

y-m

ad

em

eals

0.158

60.

124

40.

0703

0.10

030.

0292

0.165

80.1

875

-0.9

370

0.0

881

Tab

le4:

Ow

n-

and

cros

s-p

rice

elas

tici

ties

(val

ues

)

Pork

Beef

Ch

icken

Oth

er

Oth

er

Fis

h,

Eggs

Read

y-m

ad

eO

Gp

ou

ltry

meat

shellfi

shm

eals

Pork

-0.1

523

0.009

10.

0039

0.00

560.

0012

0.025

20.0

109

0.0

309

0.0

654

Beef

0.0

121

-0.0

409

0.00

120.

0016

0.00

050.

0059

0.002

40.0

073

0.0

099

Ch

icken

0.008

90.

002

0-0

.0271

0.00

150.

0007

0.004

20.0

015

0.0

052

0.0

031

Oth

er

pou

ltry

0.009

20.

001

90.

0011

-0.0

297

0.00

050.

004

50.0

017

0.0

055

0.0

053

Oth

er

meats

0.005

80.

001

60.

0014

0.00

13-0

.0180

0.00

270.

000

90.0

033

0.0

010

Fis

h,

shellfi

sh0.

024

00.

004

20.

0017

0.00

250.

0005

-0.0

861

0.005

20.0

147

0.0

332

Eggs

0.008

40.

001

40.

0005

0.00

080.

0001

0.004

2-0

.0368

0.0

052

0.0

161

Read

y-m

ad

em

eals

0.029

60.

005

20.

0021

0.00

310.

0006

0.014

80.0

065

-0.1

021

0.0

402

12

The resulting elasticities at the category level are shown in Table 3. Because pricesand market shares are different from one category to another, these elasticities do notreflect the magnitude of substitution effects among categories. We thus also provide inTable 4 elasticities in terms of market share changes (evaluated at the observed marketshare).

Cross-price elasticity values presented in Table 4 show that an increase in the priceof one category of products will increase the market shares of all other categories, es-pecially ready-made meals and vegetable food products. For the other meat categories(including beef, chicken, other poultry and other meats), the pork category benefitsthe most from an increase in the price of the considered meat category.

The second most important market share report for the meat categories is forready-made meals and then on vegetable food products, except for beef for which thereport on vegetable food products is higher. An increase in the price of the marinefood products or in the price of eggs induces higher market share reports towards thevegetable food category, the pork category and then towards the ready-made mealscategory.

Finally, regarding a change in the price of the ready-made meals, reports are mainlyin favor of the vegetable food aggregate and pork.

From the own-price elasticity results, we can thus conclude that a change in theprice of a given meat and marine product will not generate a large decrease in thepurchase of this product. These results may be explained by the existence of pre-committed demand (Tonsor and Marsh, 2007). Some purchases may be influenced bynon-price factors such as generic advertising, food safety concerns or health concerns.Other habits such as cultural or family habits could also explain this inelastic demandfor meat and marine products. Moreover, among meat categories, pork that whichbenefits most from a price increase in any other meat category. Finally, the substi-tutions with ready-made meals and vegetable food are important and may even behigher in value for the vegetable food category. This is the case when the price ofpork, marine food products or eggs increases.

4 Impact of a taxation policy on food diet

Using data on the environmental impact of our 28 differentiated meat and marineproducts, we analyse some pricing policy instruments such as taxes to infer how foodconsumption behavior could be changed towards more sustainability.

4.1 Environmental data

We use environmental data collected by Greenext, a French company specializingin calculating and analyzing the environmental impact of consumer goods. Thisdatabase contains three environmental indicators (acidification, eutrophication andclimate change) for 311 different food products. The environmental data are basedon the life-cycle analysis (LCA) of the differentiated products. LCA is a multi-stepmethod which allows us to take into account all the phases of the life-cycle of a productfrom the production of raw materials, processing, distribution and use, until the end

13

of its life.

We use this data source because it is the only source of environmental indica-tors available at such a detailed level in France that provides comparable values forthe different products. The methodology used to compute those values is consistentamong all products considered in the database, and the indicators used are the mostreliable among the environmental data actually measured. Thus, other environmentalindicators such as energy use, land use or biodiversity are not taken into account inthis study because data for these are not available in a comparable and reliable format.

Some products considered in this analysis could not be directly matched withGreenext products. For these products, we allocate the average environmental valuesof the closest Greenext products. For example, we allocate the same environmentalvalues to the three products in the beef meat category since only the values for mincedbeef are available, and because the three products are similar in terms of environmentalimpact, and they all correspond to fresh meat. In the same way, fresh chicken pieceshave the same environmental values as ready-to-cook chicken. Additionally, there isno environmental data for rabbit, thus we allocate the mean values of poultry as itscomposition is quite similar. Finally, for mixed meats, we affect the mean environ-mental values of all other meats since we could not recover the part of each meat forthis product. Regarding the vegetable food aggregate, 76% of the observations havebeen matched with their corresponding environmental values. Next, we compute themean environmental values for the three indicators, weighted by the market share ofproducts that are part of the outside good.

As shown in Table 5, beef is the meat product that has the largest environmentalimpact in terms of CO2-eq, followed by veal and lamb and sheep which are all definedas ruminant meats. The carbon impact of pork is the lowest among meat products,with GHG emissions representing roughly a third of the beef emissions. The marinefood category has the equivalent GHG emissions as pork, while eggs are even lessimpacting. Compared with meat categories, ready-made meals are less impacting asthey contain less meat per kilogram of product. Finally, as expected, vegetable foodproducts are the least impacting, with their GHG emissions representing 10% of thebeef emissions. The ranking of products regarding their impact on CO2-eq is similarfor acidification, whereas chicken and poultry are more impacting in terms of eutroph-ication.

Except for the beef category (for which we were not able to collect more preciseinformation about the environmental impact of the three beef products), we can seethat this impact is heterogeneous within the eight categories. For instance, in thepork category, the environmental indicators between blood pudding and other porkproducts double.

14

Table 5: Environmental indicators

ProductEnvironmental values (g/100g)

Acidification Eutrophication CO2-eqFresh pork 10.83 4.53 479.82Cooked ham and roasts 9.93 4.15 490.47Fresh sausages 8.87 3.70 466.53Other sausages (Strasbourg) 10.30 4.29 496.21Bacon 11.39 4.78 581.24Blood pudding 5.73 2.42 352.51Other pork 12.56 5.14 610.11Pork category 10.73 4.46 521.20Minced beef 33.72 4.62 1589.23Beef for grilling 33.72 4.62 1589.23Other beef 33.72 4.62 1589.23Beef category 33.72 4.62 1589.23Ready to cook chicken 13.89 8.24 796.74Chicken parts 13.89 8.24 796.74Other chicken 10.61 5.42 581.18Chicken category 13.06 7.53 742.47Turkey 10.56 6.00 599.87Duck 12.40 7.56 766.05Other poultry 10.23 4.61 520.25Other poultry category 10.80 5.86 606.14Veal 20.67 2.88 1148.39Lamb, sheep 28.67 3.01 1182.79Mixed meats 14.77 4.85 734.92Rabbit 10.23 4.61 520.25Other meats 14.77 4.85 734.92Other meat category 20.10 3.58 981.05Fresh fish 2.01 1.84 448.77Fresh shellfish 4.82 0.59 603.30Processed fish and shellfish 2.74 1.74 510.18Fish category 2.97 1.52 510.60Eggs 6.72 3.19 332Pizzas, quiches 3.13 1.40 268.20Snacking 5.69 1.75 414.59Ready-made meals 5.25 1.90 387.02Ready-made meals category 5.17 1.84 382.89All Inside Goods 11.92 3.77 668.37Outside Good 0.76 0.77 148.13Numbers on grey rows are weighted means

15

4.2 Simulations

Given the substitution patterns estimated in the previous section and the differences inthe environmental indicators for the various meat and marine products and a vegetablefood aggregate, an environmental tax will have differing environmental implications,depending on its design.

We consider taxes based on the CO2 equivalent emissions. Two different levels oftaxes are chosen: e56 and e200 per tonne CO2-eq. They are computed as a differ-entiated tax per kilogram of final product depending on the level of GHG emissionper product. These levels of taxes correspond respectively to a reduction in GHGemissions by 20% and 60%, that are the medium- and long-term European Union’stargeted values by 2020 and 2050. Note that these values are higher than the topprice observed on the carbon market (e30 per tonne in 2008) and much higher thanthe carbon price recently observed (e4 per tonne) (Committee on Climate Change,2014).6

Because GHG emissions are heterogeneous from one product to another (even forproducts from the same category), the impact of such a tax on prices will also differ.This heterogeneity within product categories fosters the substitutions between prod-ucts from the same category. Since beef products are the most impacting in termsof GHG emissions, their prices increase the most. A tax of e56 per tonne CO2-eqis relatively low and corresponds to an increase in the price of the product betweenroughly 2% (fish products) and 12% (beef products). A tax of e200 per tonne CO2-eqhas a more significant impact on the prices and corresponds to a rise in prices from7.2% for marine products to 42.8% for beef (Table 6).7

We consider three possible cases. First, we consider a tax on all meat and marineproducts. Because beef products are more impacting, we also consider the case whereonly beef products are taxed. Finally, we consider an intermediate case where veal,lamb and sheep products (the most impacting after beef) are taxed in addition to beef.Given the two possible levels of taxes, this leads to six possible scenarios. Table 7 sum-marizes the impacts of the tax policy on market shares for each scenario. We assumethat the market is closed, meaning that we only study the reports between categories,and that there is no variation in the total meat and marine food, and vegetable food

6In their evaluation of environmental costs, Irz et al. (2015) use values close to the price of carbonin 2008, relying on the median carbon price (e32 per tonne) estimated in a meta-analysis of the socialcost of carbon developed by Tol (2012). Our higher value is in the range of the value found for the95-percentile of the fitted distribution (e185 per tonne). Note that our assumptions on the level of thetaxes are higher compared to the values considered in Edjabou and Smed (2013), and our low valueassumption is consistent with that used in Wirsenius et al. (2011).

7Note that ready-meal products include both meat, marine and vegetable components. So, whentaxing ready-made meal products, we consider a tax on both ingredients. As we are not able torecover the proportion of meat and marine components in the final product, we simulate a scenariowith a lower tax on ready-made meals to check the robustness of our results. The level of the taxesare computed given the share of meat and marine components in the product, that is 45% for pizzas,quiches and tarts, 64% for snack food and 62% for prepared meals. We found a similar reduction inenvironmental impacts. Results are available upon request.

16

Table 6: Impact of the public policy scenarios on prices

Tax (e/ T CO2-eq / KG meat)56 200

Category Price (e/Kg) Price variation interval (%)

Pork8.80

[2.17–5.35] [7.76–19.12](0.21)

Beef10.37

[5.96–11.98] [21.30–42.80](0.55)

Chicken6.52

[3.83–8.96] [13.68–32.00](0.17)

Poultry9.66

[2.55–4.25] [9.12–15.17](1.27)

Other meats11.91

[3.16–5.59] [11.28–19.98](0.39)

Fish, shellfish10.62

[2.02–4.30] [7.23–15.36](0.58)

Eggs4.96

3.75 13.40(0.08)

Ready-made meals6.73

[2.29–3.71] [8.16–13.24](0.32)

(.) : standard deviation

The interval corresponds to the range of the tax across products within the category.

consumption. However, as we allow for an outside option composed of vegetable foodproducts, the total consumption of meat and marine products could vary and be sub-stituted with vegetable food products. When all meat and marine products are taxed,the purchase of almost all meat products decreases in favor of marine food productsand vegetable food products to a larger extent. However, when the tax is low, thedecrease in market shares is small, due to the fact that demand is quite inelastic. Thereduction is less than 1% for all categories of product except for other meats, poul-try, chicken and beef. Beef is more impacted with a decrease in its market share by7%. When the tax is much higher, the reduction in the purchases is much higher forbeef (losing more than 20% of market share), but also for other meats and chickenthat lose around 10% of their market share. The loss in the market share of meatproducts is substituted for by an increase of 4.7% of the market share for vegetablefood products and a small increase of 0.4% of the market share of marine food products.

When only beef is taxed, the market share of beef decreases by much more than inthe previous case, with a decrease of 25% with a high level of tax. Because beef can besubstituted with various meats and vegetable food products, the market shares of allother categories increase. Finally, in the intermediate case where only ruminant meatsare taxed, the market shares of all other products increase only slightly compared toa taxation on beef only.

Given the impact of different tax policies on market shares, a tax of e56 per tonneCO2-eq would lead to a very small change in GHG emissions (−1.54%) even if all meat

17

Table 7: Impact of the public policy scenarios on market shares

Tax scenarios (e/ T CO2-eq / KG meat)

All Beef Beef, veal, lamb

56 200 56 200 56 200

Category Market share (%) Market share variation(%)

Pork18.96 −0.57 −2.16 0.58 1.71 0.67 1.99

(0.66) (0.04) (0.11) (0.03) (0.09) (0.04) (0.11)

Beef4.27 −7.01 −20.38 −8.82 −25.05 −8.72 −24.75

(0.27) (0.27) (0.65) (0.30) (0.69) (0.29) (0.69)

Chicken3.12 −3.17 −10.20 0.36 1.16 0.50 1.61

(0.29) (0.06) (0.17) (0.01) (0.03) (0.02) (0.05)

Poultry3.19 −1.47 −5.08 0.45 1.39 0.57 1.78

(0.35) (0.06) (0.20) (0.02) (0.06) (0.03) (0.08)

Other meats2.09 −2.96 −9.72 0.19 0.63 −2.12 −6.83

(0.12) (0.11) (0.32) (0.01) (0.02) (0.10) (0.31)

Fish, shellfish9.02 0.21 0.4 0.60 1.75 0.68 2.01

(0.54) (0.14) (0.48) (0.03) (0.08) (0.03) (0.10)

Eggs3.51 −0.75 −2.86 0.63 1.80 0.70 2.02

(0.14) (0.08) (0.27) (0.04) (0.11) (0.05) (0.12)

Ready-made meals11.01 −0.32 −1.40 0.60 1.76 0.68 2.03

(0.80) (0.07) (0.25) (0.04) (0.10) (0.04) (0.11)

Outside Good44.83 1.41 4.68 0.20 0.49 0.20 0.51

(1.26) (0.04) (0.13) (0.02) (0.05) (0.02) (0.05)

(.) : standard deviation

18

and marine products were taxed (Table 8). Similarly, the impact on other environmen-tal indicators is limited given this small amount of tax; the highest impact being forthe acidification indicator. The environmental impact is multiplied by four with a taxof e200 per tonne CO2-eq when all products are taxed for the GHG emissions. Thecarbon tax can lead to a decrease of up to 4.8% of GHG emissions per year, 3.7% foreutrophication and 7.3% for acidification. Thus, we can see that even with a carbontax of e200 per tonne CO2-eq on all meat and marine products, we are far from thedesired carbon emission reduction of 20%.

Table 8: Environmental, nutritional and welfare indicators variations according to thepublic policy scenarios

Taxed Products All Beef Beef,veal,lamb

Tax (e/T CO2-eq / KG meat) 56 200 56 200 56 200

4 CO2-eq per year (%) −1.54 −4.77 −1.11 −3.12 −1.19 −3.36

4 Acidification (SO2) per year (%) −2.35 −7.26 −1.72 −4.83 −1.83 −5.21

4 Eutrophication (N) per year (%) −1.15 −3.71 −0.30 −0.80 −0.28 −0.75

4 Meat (per day) (Tonnes) −109.26 −362.99 −15.37 −38.09 −15.79 −39.44

4 Meat (per day and person) (g) −1.75 −5.80 −0.25 −0.61 −0.25 −0.63

4 Calories (per year and person) (%) −0.29 −0.98 −0.00 +0.01 −0.00 +0.02

4 Lipids (per year and person) (%) −0.79 −2.63 −0.01 +0.02 −0.00 +0.05

4 Proteins (per year and person) (%) −0.92 −3.00 −0.25 −0.69 −0.28 −0.78

4 SFA (per year and person) (%) −0.98 −3.22 −0.12 −0.27 −0.13 −0.30

4 Cholesterol (per year and person) (%) −1.16 −3.91 −0.01 +0.02 −0.08 −0.19

4 Welfare (%) −0.21 −0.73 −0.04 −0.12 −0.06 −0.21

Tax revenue (billion e) 1.10 3.74 0.21 0.61 0.26 0.78

Results of simulations also show that a tax on all meat and marine products leadsto a decrease in all nutritional indicators especially saturated fat acids and choles-terol.8 When only beef or ruminant meats are taxed, the impact on nutrition is almostannihilated because the reduction in the market share of beef (or beef, veal and lamb)generates a substitution, not only in the vegetable food category, but also with othermeat products that have higher nutritional indicators, such as pork. In fact, pork isthe most caloric among all categories, as it contains more saturated fat acids, morelipids and less proteins (after fish and eggs), while eggs, chicken and poultry containmore cholesterol.

8Table A in the Appendix presents the value of all nutritional indicators for each of the 28 productsand an average value for the vegetable food aggregate.

19

We can thus conclude that a tax policy targeting all meat and marine products withthe highest level of tax will be the most efficient scenario to reduce the environmentalimpact. However, Table 8 shows that if we consider consumer surplus, this scenariois not the most efficient. For the same level of tax, taxing only beef would reacharound 70% of the total environment reduction when all meat and marine productsare taxed, whereas it would deteriorate the consumer welfare five times less. Becausebeef is the most impacting meat, such a policy has the advantage of limiting thedecrease in the consumption of meat and marine products but still to substantiallylower environmental damages. This is in line with the conclusion of Wirsenius et al.(2011), that most of the effect of a GHG tax on animal food can be captured by taxingthe consumption of ruminant meat alone. This scenario would be the best policy as itis an acceptable trade off between reducing the environmental impact and not affectingconsumer welfare too much.

5 Conclusion and discussion

Sustainable diet recommendations encourage a reduction in meat consumption. Whilethe results in the literature show that a reduction in meat consumption would effec-tively have positive benefits both in terms of environment and health (Irz et al. (2015)and Soret et al. (2014)), there is also a consensus to say that consumers’ dietary habitsare difficult to change (Perez-Cueto et al., 2013) through public information interven-tions. This paper analyzes a public policy which targets a change in the environmentof consumers: a tax policy on meat and marine products. We address this questiondealing with two different intervention strategies: first, to change the global consump-tion habits toward all meat and marine products by reducing the consumption of thewhole meat and marine products category; or second, to favor changes within this cat-egory and then promote substitutions towards meat and marine products which havea lower environmental impact.

Our analysis of French meat and marine consumption has contributed to the policydebate regarding the ability of environmental taxes to mitigate environmental dam-ages linked to animal product consumption. Our results, based on a flexible demandmodel for meat and marine products disaggregated at a very detailed category level,initially confirm that it is indeed difficult to significantly decrease the market sharesof meat and marine products even with a high level of taxes. The reason is that thedemand for meat and marine products is quite inelastic. A change in the price of meatand marine products generates quite low and partial substitutions with vegetable foodproducts. This is because part of the substitutions occur within the meat and marineproduct categories; pork being the meat product that benefits most from a rise in theprice of other products. However, if a relatively low level of tax fails to significantlyreduce the different environmental damage indicators, our results suggest that a highlevel of tax (corresponding to the long-term European Union’s commitments on carbonprices) will have a significant impact on the environmental footprint: almost 5% forGHG emissions and up to 7% for acidification. This suggests the implementation of ahigh level of tax in order to mitigate the environmental impact. Moreover, in a contextwhere the prevalence of obesity is a major public health concern in most developedcountries, it is essential to confirm that environmental and nutritional policies head inthe same direction. We show that such taxes will not damage the nutritional quality

20

of food consumption.

Finally, an interesting outcome from our simulations of tax policies suggests thateven if we only tax the main environmentally-impacting meat and marine products(beef meat), we may recover a large part of the environmental benefits without sig-nificantly hurting consumers. Such an instrument will only partially reduce the con-sumption of meat, as part of the substitution will occur within the meat and marineproduct categories. Thus, it is better from a welfare point of view to design a policythat targets only beef and not all meat and marine categories.

Our analysis is a first attempt to better include the impact of consumers’ purchasingbehaviors on the design of an environmental tax policy. To go further, there is alsoa need to better anticipate the change in the behaviors of the meat supply chainwhen facing a tax policy, and such an analysis is important to better evaluate theenvironmental impacts. Indeed, Bonnet and Requillart (2013) show that, in the caseof public health policy, ignoring firms’ strategic pricing decisions leads to a biaisedeffect in estimating the impact of a sugar tax on the soft drink industry. However, theanalysis of the supply chain reaction would require more precise information about thesupply chain structure and decision variables at each stage of the supply chain. It isthus beyond the scope of this paper, even though economists should be encouraged toput this issue on their agenda for future work.

21

Appendices

Table A: Nutritional indicators

ProductNutritional values

Calories Lipids Proteins SFA Cholesterol(kcal/100g) (g/100g) (g/100g) (g/100g) (mg/100g)

Fresh pork 166.71 9.70 19.82 3.76 54.36Cooked ham and roasts 116.75 3.86 19.73 0.90 42.43Fresh sausages 296.60 25.00 16.91 9.83 70.33Other sausages (Strasbourg) 304.44 27.80 12.50 9.94 66.20Bacon 260.81 21.07 17.28 7.88 57.58Blood pudding 253.70 20.37 15.71 10.22 137.05Other pork 335.60 27.33 21.24 10.08 89.70Pork category 236.55 17.44 19.11 6.53 67.77Minced beef 189.91 12.48 19.22 5.30 65.55Beef for grilling 151.53 5.00 26.63 1.89 57.07Other beef 173.77 8.60 22.60 3.28 92.79Beef category 225.06 11.46 29.87 4.63 91.34Ready to cook chicken 231.00 14.20 25.90 4.20 122.00Chicken parts 167.22 7.84 23.92 2.28 88.78Other chicken 183.75 9.86 22.66 3.05 92.65Chicken category 192.64 10.47 24.26 3.11 100.83Turkey 170.21 5.88 25.70 1.91 74.47Duck 174.57 7.87 24.84 3.06 171.76Other poultry 324.97 27.52 17.53 10.44 343.15Other poultry category 217.39 12.73 23.09 4.67 172.65Veal 222.38 14.37 23.00 6.84 113.20Lamb, sheep 231.73 15.80 20.74 5.63 134.85Mixed meats 250.55 18.80 19.57 7.02 85.58Rabbit 192.98 12.14 20.22 5.08 86.88Other meats 229.83 15.38 22.23 4.89 88.97Other meat category 225.23 15.12 21.49 6.13 109.67Fresh fish 134.52 5.37 19.10 1.00 53.54Fresh shellfish 101.14 2.29 16.99 0.40 83.40Transformed fish and shellfish 160.62 8.13 19.03 1.80 62.94Fish category 138.55 5.90 18.60 1.22 64.40Eggs 142.16 9.88 12.60 2.64 378.00Pizzas, quiches 227.77 9.63 9.99 3.73 25.49Snacking 215.83 10.31 9.58 4.04 29.21Ready-made meals 156.46 7.70 9.00 2.79 39.69Ready-made meals category 210.31 11.01 12.72 3.79 144.13All Inside Goods 190.45 11.42 18.11 4.05 91.15Outside Good 126.43 2.99 4.09 0.79 3.90

The study uses “NutriXConso” database (Kantar home scan data linked with Nutritional Data),

cf (De Mouzon and Orozco, 2011)

22

Bibliography

Afssa (2009). Etude individuelle nationale des consommations alimentaires 2 (inca 2)2006-2007. Technical report, Agence Francaise de Securite Sanitaire des Aliments.

Austgulen, M. (2014). Environmentally sustainable meat consumption: An analysis ofthe norwegian public debate. Journal of Consumer Policy, 37(1):45 – 66.

Bonnet, C. and Requillart, V. (2013). Tax incidence with strategic firms on the softdrink market. Journal of Public Economics, 106:77–88.

Committee on Climate Change (2014). Meeting carbon budgets 2014 : Progress reportto parliament. Technical report, Committee on Climate Change.

De Mouzon, O. and Orozco, V. (2011). Nutrixconso : recherche et appariement dedonnees d’achats et de donnees nutritionnelles. Cahier des Techniques de l’INRA,(74):42 – 83.

Dong, D., Davis, C. G., and Stewart, H. (2015). The quantity and variety of households’meat purchases: A censored demand system approach. Agricultural Economics,46(1):99 – 112.

Edjabou, L. D. and Smed, S. (2013). The effect of using consumption taxes on foodsto promote climate friendly diets–the case of denmark. Food Policy, 39:84–96.

Feng, K., Hubacek, K., Guan, D., Contestabile, M., Minx, J., and Barrett, J. (2010).Distributional effects of climate change taxation: the case of the uk. Environmentalscience & technology, 44(10):3670–3676.

Fiala, N. (2008). Meeting the demand: An estimation of potential future greenhousegas emissions from meat production. Ecological Economics, 67(3):412–419.

FranceAgriMer (2015). Consommation des produits carnes en 2014. Technical report,FranceAgriMer.

Gallet, C. A. (2010). Meat meets meta: a quantitative review of the price elasticity ofmeat. American Journal of Agricultural Economics, 92(1):258–272.

Hallstrom, E., Roos, E., and Borjesson, P. (2014). Sustainable meat consumption: Aquantitative analysis of nutritional intake, greenhouse gas emissions and land usefrom a swedish perspective. Food Policy, 47:81 – 90.

Hedenus, F., Wirsenius, S., and Johansson, D. (2014). The importance of reducedmeat and dairy consumption for meeting stringent climate change targets. ClimaticChange, 124(1/2):79 – 91.

Henchion, M., McCarthy, M., Resconi, V. C., and Troy, D. (2014). Meat consumption:Trends and quality matters. Meat Science, 98(3):561 – 568. Meat Science, Sustain-ability & Innovation : 60th International Congress of Meat Science and Technology17-22 August 2014, Punta del Este, Uruguay.

Horrigan, L., Lawrence, R. S., and Walker, P. (2002). How sustainable agriculturecan address the environmental and human health harms of industrial agriculture.Environmental health perspectives, 110(5):445.

23

Irz, X., Leroy, P., Requillart, V., and Soler, L.-G. (2015). Welfare and sustainabilityeffects of dietary recommendations. TSE Working Papers 15-565, Toulouse Schoolof Economics (TSE).

McMichael, A. J., Powles, J. W., Butler, C. D., and Uauy, R. (2007). Food, livestockproduction, energy, climate change, and health. Lancet, 370(9594):1253 – 1263.

Nevo, A. (2001). Measuring market power in the ready-to-eat cereal industry. Econo-metrica, 69:307–342.

Perez-Cueto, F. J., Skov, L. R., and Mikkelsen, B. E. (2013). Nutrition labelling,environment, sustainability. Public health nutrition, 16(10):1908–1909.

Quinet, A. (2009). La valeur tutelaire du carbone, rapport du centre d’analysestrategique. La documentation Francaise.

Revelt, D. and Train, K. (1998). Mixed logit with repeated choices: householdschoicesof appliance efficiency level. Review of Economics & Statistics, 80(4):647 – 657.

Reynolds, C. J., Piantadosi, J., Buckley, J. D., Weinstein, P., and Boland, J. (2015).Evaluation of the environmental impact of weekly food consumption in differentsocio-economic households in australia using environmentally extended input–outputanalysis. Ecological Economics, 111:58–64.

Rios-Nunez, S. M., Coq-Huelva, D., and Garcia-Trujillo, R. (2013). The spanish live-stock model: A coevolutionary analysis. Ecological Economics, 93:342 – 350.

Smed, S., Jensen, J. D., and Denver, S. (2007). Socio-economic characteristics and theeffect of taxation as a health policy instrument. Food Policy, 32(5):624–639.

Soret, S., Mejia, A., Batech, M., Jaceldo-Siegl, K., Harwatt, H., and Sabate, J. (2014).Climate change mitigation and health effects of varied dietary patterns in real-life settings throughout north america. The American journal of clinical nutrition,100(Supplement 1):490S–495S.

Tol, R. S. (2012). A cost–benefit analysis of the eu 20/20/2020 package. Energy Policy,49:288–295.

Tonsor, G. T. and Marsh, T. L. (2007). Comparing heterogeneous consumption in u.s.and japanese meat and fish demand. Agricultural Economics, 37(1):81 – 91.

Traill, W., Mazzocchi, M., Niedzwiedzka, B., Shankar, B., and Wills, J. (2013). Theeatwell project: recommendations for healthy eating policy interventions across eu-rope. Nutrition Bulletin, 38(3):352–357.

Tukker, A., Goldbohm, R. A., de Koning, A., Verheijden, M., Kleijn, R., Wolf, O.,Perez-Dominguez, I., and Rueda-Cantuche, J. M. (2011). Environmental impacts ofchanges to healthier diets in europe. Ecological Economics, 70(10):1776 – 1788.

UNEP, U. N. E. P. (2010). A report of the working group on the environmental im-pacts of products and materials to the international panel for sustainable resourcemanagement. UNEP Paris.

24

Vinnari, M. and Tapio, P. (2012). Sustainability of diets: From concepts to governance.Ecological Economics, 74:46–54.

Wirsenius, S., Hedenus, F., and Mohlin, K. (2011). Greenhouse gas taxes on ani-mal food products: rationale, tax scheme and climate mitigation effects. ClimaticChange, 108(1-2):159–184.

25